Power Transformer leakage

[BudP]

The range of twists per inch has influence upon what type of signals you are rejecting. The maximum twists per inch limit is an attempt to reject high frequency electrostatic noise. The low number of turns per inch keeps you above the ELF coupling of mains and rectifier emitted flux fields.

From practical experience this rather narrow range has the highest percentage of corrective effect. Does not mean that you cannot twist filaments leads as tightly as possible and others leads much more loosely than I am suggesting. I have had good success with both extremes, just that there is a middle ground that seems to be most productive of benefit.

I suspect there are two converging events addressed. One is the use of periodic shielding of one conductive plate by another, in a nodal length that disrupts the electrostatic field coupling Certainly the E Field D Field relationships are severely messed with by the imposition of differing conductive plates and their surrounding dielectrics. I do not recall ever reading a systematic study on this issue though, so I can only present empirically derived information.

The second is to cause the potential antenna wires to cut the flux lines in a nodal rhythm that deconstructs the current that would normally arise as an induced "information packet" that would be understood as a "hum". Certainly the electrons are still being driven by the external field event but the information content is rendered into incoherent noise. Twisting the emitter cables in the mentioned range along with the receiver cables has, again, brought the most satisfactory results, in my own experience, in trouble shooting hum in other peoples products.

I am quite certain that no hard rules apply and certainly any attempt at a mathematical modeling of these two field events, within any given amplifier and it's physical layout, is going to take most of a lifetime. I am also certain that many other solutions, from many other empirical experiments are just as valid as mine.

I can strongly recommend at least a casual brush with Ralph Morrison's book "Grounding and Shielding Techniques" fourth edition ISBN 0-471-24518-6. While he does not address this particular question directly, his explanations and pictorial representations of field events and their coupling modes and vectors is exceptionally clear and educational.

Bud

[audio_moksha]

Sorry for the late response folks for it is alomst pring break here at Whitman College and my professors are overloading me with work. Anyway, I twisted the filament wires and even tried moving the transformer around but to no effect.

I then sandwiched a steel square between tow aluminium squares and I put them between the transformers and it definitely helped. Now I have to put my ears against the drivers and really listen hard to hear the humm. Ofcourse it would have been good if there was no hum at all.

The power transformer is really noisy though; can be heard a foot away form the amp. It does not get hot but even then the noise is annoying. Perhaps buying a new power transformer may be the answer.

[EC8010]

I don't think you've posted a diagram of your HT power supply, there might be something there that is causing the transformer problem...

[ulibub]

Originally posted by audio_moksha

ulibub, the hum sounds like it might be 60Hz or 120Hz hum its not really buzzy and I can only hear it when I have my ear against the speaker but the perfectionist in me wants the amp to be completely quiet. Also the power transformer is mechanically buzzy but it does not get hot at all.

...then I still can assume the following:

1. poor quality transformer. This mechanical hum can also couple into the tubes and can get amplified by their microphony. Maybe that's the hum you hear in your speakers. Transformers of that size have to be mechanically quiet, especially if not fully loaded.

2. did you check your mains voltage? how "dirty" is it? If it has high amounts of harmonics and non-sinusoidal signals, it is VERY difficult to get this noise out - and usually they give some buzzing sound. In that way, a well performing mains filter might help.... My big Circlotron amp , which still isn't as "dead" quiet as I'd like it, also shows its buzz depending on daytime... - often typical for this kind of interference.

Uli

[BudP]

Power transformers that feed a rectifier are being asked to provide current pulses, very narrow in width in a time domain and so quite strong in the amplitude domain. Some circuits force 100 amp pulses for just a 40 watt amplifier, due to the amount of filter capacitance in the rectifier circuit. A FWCT will usually be worse in this regard than a FWB. If your amp uses rectifier tubes then it is almost certainly a FWCT.

If you decide to change your transformer, then a larger unit, with considerably less flux density in the core, and a FWB constructed of HexFred soft recovery diodes will help with the current spikes. Ypu might also look at the power supply in Lynn Olson's latest Karna issue. This is a public domain design so you may copy it at will. Notice the very small lead capacitor he uses, 1.5 mfd, this allows the power transformer to have a nearly sinusoidal current draw with a reduction in physical noise. With a brute force power transformer, one with less than 3% voltage drop from no load to full load, you can reduce the power supply capacitence to a minimum and speed up your amps response to transient events. Adding a signal return to ground buss with either Litz wire or a very large diameter stranded wire, with a plastic sleeve (Nylon selvage for the Litz) will also help greatly.

The only way to kill the mechanical noise in your current unit is to pull the end bells and find a transformer company to vacume impregnate the unit and bake it solid again. Asking them to wedge the core material at the winding form will also help.

My personal solution would be to buy the brute force transformer, though that is often scorned by designers as having no finesse. Just means they were not often allowed to spend the bill of materials money on expensive transformers and so had to find a solution, usually using banks of much less expensive capacitors and voltage regulators for power storage. Nothing wrong with either approach but the brute force method will net you better music.

Bud

[BudP]

The test you made with the piece of steel buffered by aluminum does show that it is mostly external field linking that you have to solve. The thing to keep in mind is that these things are essentially infinite in reach so you have only a few methods of control open to you.

One is just what you did, create a flux shadow by compressing the field lines directly between the two transformers by attaching them to a mildly permeable material and steering them around the OPT's. More of this is a good thing, to a point. Extending the shield piece too far beyond the core pieces will actually hurt your effort. so do not get much more than an inch beyond the transformers metal work. You do not need to shield this external sheet with anything more than .010 of cardboard, even a piece of scotch tape is a good enough insulator. In some cases attaching the shield metal directly to the core of the emitter is a good idea, you will not short the core out.

As an aside, when wrapping a core band around a transformer core, it should extend about 0.050" beyond either core edge. Less will increase the emitted field dramatically and more will be wasted to no further benefit.

When reorienting the power do not hesitate to also reorient the OPT's and do not worry too much about getting everything nicely lined up and squared away. Your situation may require all three to be at odd angles to each other and the edges of the chassis. Also it would not hurt to alternately flop them over on their sides, if that is possible, and steer the power generated field around with the two different types of transformers alternately on their sides and then on end bell faces.

These two, the shield and diverse core orientations, may also need to have the OPT's on standoff's to complete the process.

Good luck, keep track of what changes made what sorts of improvements and just as importantly, what made it worse. If you can report your findings it will aid many others.

Bud

[ulibub]

...I never liked abreviations...

Please tell me - what's a FWCT and whats a FWB????

Uli

[EC8010]

I think BudP means Full Wave Centre-Tapped and Full Wave Bridge, but if so, I don't see that one should draw higher ripple current than the other unless we're down at the point where the additional series resistance of the extra diode limits current.

[audio_moksha]

Hello gentlemen,
I will probably post my schematics later this afternoon. Anyway to answer some of your questions:
i have a feeling that ulibub is right: the transformer sucks. I know for sure that the hum is magnetic leakage directly into the OPt's because the hum comes on instantly without the tubes even warming up and it also disappears immediately. I have tried plugging it in various friends' houses with the same problem so I don't think its my mains voltage. I was also hanging out with Jason Flanary of Cain and Cain and I played my amp on the Cain Ben ES model and the hum was almost undectectable. Do you thing the sensitivity of the driver plays a big role in this? (I have a BR Hemp acoustic FR 8 drivers w/95dB eff.)

BudP, right now I have not insulated the aluminium sheets; do you think using a cardboard insulating sheet would help? Also i already am using hexfred soft start diodes in FWCT topology. i'll post my schematic as soon as I can. also i'm going to try and flop the OPT's on their sides and I'll et you know if things are.

Thanks everyone!

[BudP]

I suspect the problem is due to my usual vague mental wanderings.

The angle of attack and height are not the issue, rather it appears to be related to the actual current draw period and the rate of decline in that draw. I have attached a zip of two PSUD files showing the power circuits for the Karna amplifier. One is the original Full Wave Center Tap and the other the latest Full Wave Bridge, though the rectifier for that one has now migrated to a hybrid, with Hexfreds on the minus side of thee bridge.

After loading the files in separate copies of PSUD run the simulations and then window in on the saw tooth of IT1. That relatively gentle angle of current decline in the Full Wave Bridge circuit does appear to result in a significant decline in power growl from these transformers. Both units were vacume impregnated with a material that mimics obsidian when cured out and both were wedged and have the same flux density. Power growl is not buzz, as that comes from loose core flanges, usually from the E /I center tongue, down near the middle of the stack, where an incomplete penetration of potting material will allow movement. This growl is the transformer trying to deal with the current draw and crest factors, between it and the rectifier circuit. The growl mechanism is of course from the flux, induced by current flow and the consequent magnetostriction of the core sheaves, as they literally swell and shrink and bend and twist due to their flux manipulation by the current.

I will note that a more conventional lead capacitor, in place of the one chosen by Lynn, tends to negate this difference, even though there does still seem to be a physical noise benefit to using the Full Wave Bridge circuit. There is also the voltage/corona stress on coil dielectrics and anything you can do to reduce physical abuse and voltage stress is going to lengthen the life of this sorely abused component

Bud